Efficient retrieval of external knowledge bases and web pages is crucial for enhancing the reasoning abilities of LLMs. Previous works on training LLMs to leverage external retrievers for solving complex problems have predominantly employed end-to-end reinforcement learning. However, these approaches neglect supervision over the reasoning process, making it difficult to guarantee logical coherence and rigor. To address these limitations, we propose Thinker, a hierarchical thinking model for deep search through multi-turn interaction, making the reasoning process supervisable and verifiable. It decomposes complex problems into independently solvable sub-problems, each dually represented in both natural language and an equivalent logical function to support knowledge base and web searches. Concurrently, dependencies between sub-problems are passed as parameters via these logical functions, enhancing the logical coherence of the problem-solving process. To avoid unnecessary external searches, we perform knowledge boundary determination to check if a sub-problem is within the LLM's intrinsic knowledge, allowing it to answer directly. Experimental results indicate that with as few as several hundred training samples, the performance of Thinker is competitive with established baselines. Furthermore, when scaled to the full training set, Thinker significantly outperforms these methods across various datasets and model sizes.

The open nature of wireless communications renders unmanned aerial vehicle (UAV) communications vulnerable to impersonation attacks, under which malicious UAVs can impersonate authorized ones with stolen digital certificates. Traditional fingerprint-based UAV authentication approaches rely on a single modality of sensory data gathered from a single layer of the network model, resulting in unreliable authentication experiences, particularly when UAVs are mobile and in an openworld environment. To transcend these limitations, this paper proposes SecureLink, a UAV authentication system that is among the first to employ cross-layer information for enhancing the efficiency and reliability of UAV authentication. Instead of using single modalities, SecureLink fuses physical-layer radio frequency (RF) fingerprints and application-layer micro-electromechanical system (MEMS) fingerprints into reliable UAV identifiers via multimodal fusion. SecureLink first aligns fingerprints from channel state information measurements and telemetry data, such as feedback readings of onboard accelerometers, gyroscopes, and barometers. Then, an attention-based neural network is devised for in-depth feature fusion. Next, the fused features are trained by a multi-similarity loss and fed into a one-class support vector machine for open-world authentication. We extensively implement our SecureLink using three different types of UAVs and evaluate it in different environments. With only six additional data frames, SecureLink achieves a closed-world accuracy of 98.61% and an open-world accuracy of 97.54% with two impersonating UAVs, outperforming the existing approaches in authentication robustness and communication overheads. Finally, our datasets collected from these experiments are available on GitHub: https://github.com/PhyGroup/SecureLink_data.

Feature selection is a critical stage in intrusion detection, aiming to eliminate irrelevant and redundant features while retaining those with higher discriminative power, thereby enabling more accurate and cost-efficient detection. Existing methods mainly include global feature selection methods and class-specific feature selection methods. The former generate a unified feature subset shared by all attack categories, making it difficult to reveal the differentiated feature patterns relied upon by different attack categories. Although the latter can construct category-level feature subsets separately, the union of multiple subsets often increases the final number of selected features. Meanwhile, both types of methods rely on hybrid internal–external features of information systems whose number tends to be virtually unlimited, and thus still face high overall feature usage cost in real-world deployment. To address the above issues, this paper proposes an attack category mapping modeling method based on system-level features for intrusion detection systems. Based on a finite set of system-level features, the method constructs a mapping feature group for each attack category by combining correlation quantification measures specifically designed for intrusion detection scenarios with mapping rules, and then generates the final mapping model through consolidation. The proposed method can reveal the structured correspondence between different attack categories and their key system-level feature groups, thereby enhancing the interpretability of intrusion detection results and reducing feature usage cost. Comparative experiments were conducted against existing feature selection methods for intrusion detection on the NSL-KDD, UNSW-NB15, and CIC-IDS2017 datasets using three classifiers: C4.5 decision tree, Random Forest, and XGBoost. The experimental results show that the proposed method achieves performance comparable to existing methods across multiple detection metrics, while reducing the numbers of selected features on the three datasets to 20, 24, and 21, respectively, which are substantially fewer than those selected by most compared methods.

Due to the openness of the wireless medium, smartphone users are susceptible to user privacy attacks, where user privacy information is inferred from encrypted Wi-Fi wireless traffic. Existing attacks are limited to recognizing mobile apps and their actions and cannot infer the smartphone user identity, a fundamental part of user privacy. To overcome this limitation, we propose U-Print, a novel attack system that can passively recognize smartphone apps, actions, and users from over-the-air MAC-layer frames. We observe that smartphone users usually prefer different add-on apps and in-app actions, yielding different changing patterns in Wi-Fi traffic. U-Print first extracts multi-level traffic features and exploits customized temporal convolutional networks to recognize smartphone apps and actions, thus producing users' behavior sequences. Then, it leverages the silhouette coefficient method to determine the number of users and applies the k-means clustering to profile and identify smartphone users. We implement U-Print using a laptop with a Kali dual-band wireless network card and evaluate it in three real-world environments. U-Print achieves an overall accuracy of 98.4% and an F1 score of 0.983 for user inference. Moreover, it can correctly recognize up to 96% of apps and actions in the closed world and more than 86% in the open world.

The rapid advancement of technologies such as virtual and augmented reality has garnered substantial attention from both academia and industry. As the foundation of immersive multimedia content, omnidirectional image generation necessitates the development of robust and efficient image stitching algorithms. Unlike planar image stitching, omnidirectional image stitching involves stitching images captured by binocular or quadrinocular camera systems, introducing lager parallaxes (e.g., 90° or even 180°), which lead to pronounced distortions and hard-to-remove artifacts. Conventional planar image stitching methods based on optical flow typically employ unidirectional models. However, the intrinsic spatial relationships between sub-images in omnidirectional image necessitate the use of bidirectional optical flow. Existing approaches often estimate optical flow for each branch independently through simple feature mapping, neglecting the latent correlations between bidirectional flow. To address these challenges, we propose using a seam-driven approach to replace the traditional weighted blending strategy to effectively minimize artifacts. Additionally, we incorporate an advanced attention mechanism to establish a "bridge" that enables joint optimization of the two optical flow branches. Finally, we lightweight the model at a small cost of precision. Experimental results demonstrate that our method comprehensively outperforms the baseline model and generates natural omnidirectional images.

Radio frequency (RF) fingerprinting techniques provide a promising supplement to cryptography-based approaches but rely on dedicated equipment to capture in-phase and quadrature (IQ) samples, hindering their wide adoption. Recent advances advocate easily obtainable channel state information (CSI) by commercial WiFi devices for lightweight RF fingerprinting, while falling short in addressing the challenges of coarse granularity of CSI measurements in an open-world setting. In this paper, we propose CSI2Q, a novel CSI fingerprinting system that achieves comparable performance to IQ-based approaches. Instead of extracting fingerprints directly from raw CSI measurements, CSI2Q first transforms frequency-domain CSI measurements into time-domain signals that share the same feature space with IQ samples. Then, we employ a deep auxiliary learning strategy to transfer useful knowledge from an IQ fingerprinting model to the CSI counterpart. Finally, the trained CSI model is combined with an OpenMax function to estimate the likelihood of unknown ones. We evaluate CSI2Q on one synthetic CSI dataset involving 85 devices and two real CSI datasets, including 10 and 25 WiFi routers, respectively. Our system achieves accuracy increases of at least 16% on the synthetic CSI dataset, 20% on the in-lab CSI dataset, and 17% on the in-the-wild CSI dataset.

The proliferation of time-sensitive Internet of Things (IoT) applications has significantly increased the demand for real-time communication in uplink orthogonal frequency division multiple access (UL-OFDMA) Wi-Fi networks. Despite extensive studies, how to meet the heterogeneous age of information (AoI) requirements across different stations (STAs) in Wi-Fi networks remains an open question. To tackle this issue, we propose a multi-agent reinforcement learning (MARL) resource allocation algorithm based on independent hybrid proximal policy optimization (IHPPO), aiming to minimize the AoI and power consumption for each STA while guaranteeing the heterogeneous AoI requirements among STAs within a resource-constrained environment. Specifically, the proposed strategy utilizes HPPO to directly optimize the original hybrid action space by combining discrete resource unit (RU) selection and continuous transmit power adjustment. Extensive simulations demonstrate the superior performance of the IHPPO mechanism in terms of convergence performance and the trade-off between AoI and power consumption, relative to the decomposed multi-agent deep deterministic policy gradient (DE-MADDPG) algorithm, the fully decentralized MADDPG (FD-MADDPG) algorithm, and the random method in different access scenarios.

Due to the high fatality rate of cancer, timely detection and treatment in the early stage of cancer is very important. In this study, a method for constructing gene mutation maps based on the principle of RGB three-channel image was proposed to realize the dimensional transformation of somatic mutation data, making it suitable for the current image classification model. In order to better capture the global features of the mutation map, this paper proposes a network model M-MNet based on the inverted residual module and the multi-head self-attention module, which can effectively extract both local and global features, and the classification effect is better than that of the existing network models.

Many distributed training techniques like Parameter Server and AllReduce have been proposed to take advantage of the increasingly large data and rich features. However, stragglers frequently occur in distributed training due to resource contention and hardware heterogeneity, which significantly hampers the training efficiency. Previous works only address part of the stragglers and could not adaptively solve various stragglers in practice. Additionally, it is challenging to use a systematic framework to address all stragglers because different stragglers require diverse data allocation and fault-tolerance mechanisms. Therefore, this paper proposes a unified distributed training framework called AntDT (Ant Distributed Training Framework) to adaptively solve the straggler problems. Firstly, the framework consists of four components, including the Stateful Dynamic Data Sharding service, Monitor, Controller, and Agent. These components work collaboratively to efficiently distribute workloads and provide a range of pre-defined straggler mitigation methods with fault tolerance, thereby hiding messy details of data allocation and fault handling. Secondly, the framework provides a high degree of flexibility, allowing for the customization of straggler mitigation solutions based on the specific circumstances of the cluster. Leveraging this flexibility, we introduce two straggler mitigation solutions, namely AntDT-ND for non-dedicated clusters and AntDT-DD for dedicated clusters, as practical examples to resolve various types of stragglers at Ant Group. Justified by our comprehensive experiments and industrial deployment statistics, AntDT outperforms other SOTA methods more than 3 × in terms of training efficiency. Additionally, in Alipay's homepage recommendation scenario, using AntDT reduces the training duration of the ranking model from 27.8 hours to just 5.4 hours.

Large-scale sequencing data is used by biologists to understand the biological systems and molecular mechanisms of disease. One challenge is to effectively use valuable information from different cancer omics to produce more accurate and reliable subtypes. We propose a clustering strategy based on three correlation matrices (CTCM) to identify cancer subtypes in multi-omics data. Connectivity matrix, similarity matrix and resampling matrix collect useful data information from different perspectives. The connection matrix divides the raw data into stable subtypes by adding noise to simulate the systematic error of the sequencing platform. The similar matrix use Gaussian kernel functions to construct connections between samples as the "skeleton" of the whole. The resampling matrix adapts to the explosive growth of data by sampling subsets. For each omics data, we combine the connectivity matrix and resampling matrix with the similarity matrix to generate an iterative version. Iterating the three relationship matrices for each omics produces a fusion matrix that is used for spectral clustering to identify cancer subtypes. Compared with six other state-of-the-art multi-omics clustering methods, CTCM achieves excellent performance on the benchmark data sets of simulation and TCGA databases. The method is general enough to replace existing unsupervised clustering techniques outside the scope of biomedical research to integrate multiple types of data.

With the rapid development of Graph Neural Networks (GNNs), more and more studies focus on system design to improve training efficiency while ignoring the efficiency of GNN inference. Actually, GNN inference is a non-trivial task, especially in industrial scenarios with giant graphs, given three main challenges, i.e., scalability tailored for full-graph inference on huge graphs, inconsistency caused by stochastic acceleration strategies (e.g., sampling), and the serious redundant computation issue. To address the above challenges, we propose a scalable system named InferTurbo to boost the GNN inference tasks in industrial scenarios. Inspired by the philosophy of "think-like-a-vertex", a GAS-like (Gather-Apply-Scatter) schema is proposed to describe the computation paradigm and data flow of GNN inference. The computation of GNNs is expressed in an iteration manner, in which a vertex would gather messages via in-edges and update its state information by forwarding an associated layer of GNNs with those messages and then send the updated information to other vertexes via out-edges. Following the schema, the proposed InferTurbo can be built with alternative backends (e.g., batch processing system or graph computing system). Moreover, InferTurbo introduces several strategies like shadow-nodes and partial-gather to handle nodes with large degrees for better load balancing. With InferTurbo, GNN inference can be hierarchically conducted over the full graph without sampling and redundant computation. Experimental results demonstrate that our system is robust and efficient for inference tasks over graphs containing some hub nodes with many adjacent edges. Meanwhile, the system gains a remarkable performance compared with the traditional inference pipeline, and it can finish a GNN inference task over a graph with tens of billions of nodes and hundreds of billions of edges within 2 hours.

Sparse vector coding (SVC) is emerging as a key enabler for short-packet ultrareliable and low-latency communications (URLLCs), since it displays good block error rate (BLER) performance and can achieve low transmission latency. In this article, we propose an SVC-based sparse superimposed transmission (SVC-ST) coding scheme to further enhance the BLER performance of the SVC scheme. At the encoding side, a portion of transmission bits is represented by nonzero position indices of the sparse vector. The remaining bits are equally split into multiple streams and then mapped into the nonzero positions of sparse vector via quadrature amplitude modulation (QAM) with constellation rotation (CR). We afterward adopt the multipath matching pursuit-based soft decoding (MMP-SD) to recover the transmission packet. The BLER and bit error rate (BER) analyses of the SVC-ST scheme demonstrate the validity and rationality of our study. Moreover, we find from the simulation results that the proposed SVC-ST scheme outperforms SVC and its enhanced version (ESVC) schemes in terms of BLER and latency performance.

Sparse code multiple access (SCMA), which helps improve spectrum efficiency (SE) and enhance connectivity, has been proposed as a non-orthogonal multiple access (NOMA) scheme for 5G systems. In SCMA, codebook design determines system overload ratio and detection performance at a receiver. In this paper, an SCMA codebook design approach is proposed based on uniquely decomposable constellation group (UDCG). We show that there are N+1 ( N ≥ 1) constellations in the proposed UDCG, each of which has M (M ≥ 2) constellation points. These constellations are allocated to users sharing the same resource. Combining the constellations allocated on multiple resources of each user, we can obtain UDCG-based codebook sets. Bit error ratio (BER) performance will be discussed in terms of coding gain maximization with superimposed constellations and UDCG-based codebooks. Simulation results demonstrate that the superimposed constellation of each resource has large minimum Euclidean distance (MED) and meets uniquely decodable constraint. Thus, BER performance of the proposed codebook design approach outperforms that of the existing codebook design schemes in both uncoded and coded SCMA systems, especially for large-size codebooks.

Machine learning over graphs has been emerging as powerful learning tools for graph data. However, it is challenging for industrial communities to leverage the techniques, such as graph neural networks (GNNs), and solve real-world problems at scale because of inherent data dependency in the graphs. As such, we cannot simply train a GNN with classic learning systems, for instance, parameter server that assumes data parallelism. Existing systems store the graph data in-memory for fast accesses either in a single machine or graph stores from remote. The major drawbacks are three-fold. First, they cannot scale because of the limitations on the volume of the memories, or the bandwidth between graph stores and workers. Second, they require extra development of graph stores without well exploiting mature infrastructures such as MapReduce that guarantee good system properties. Third, they focus on training but ignore optimizing the performance of inference over graphs, thus makes them an unintegrated system. In this paper, we design AGL, a scalable and integrated system, with fully-functional training and inference for GNNs. Our system design follows the message passing scheme underlying the computations of GNNs. We design to generate the K -hop neighborhood, an information-complete subgraph for each node, as well as do the inference simply by merging values from in-edge neighbors and propagating values to out-edge neighbors via MapReduce. In addition, the K -hop neighborhood contains information-complete subgraphs for each node, thus we simply do the training on parameter servers due to data independence. Our system AGL, implemented on mature infrastructures, can finish the training of a 2-layer GNN on a graph with billions of nodes and hundred billions of edges in 14 hours, and complete the inference in 1.2 hours.

Link prediction is widely used in a variety of industrial applications, such as merchant recommendation, fraudulent transaction detection, and so on. However, it's a great challenge to train and deploy a link prediction model on industrial-scale graphs with billions of nodes and edges. In this work, we present a scalable and distributed framework for semi-supervised link prediction problem (named DSSLP), which is able to handle industrial-scale graphs. Instead of training model on the whole graph, DSSLP is proposed to train on the k-hops neighborhood of nodes in a mini-batch setting, which helps reduce the scale of the input graph and distribute the training procedure. In order to generate negative examples effectively, DSSLP contains a distributed batched runtime sampling module. It implements uniform and dynamic sampling approaches, and is able to adaptively construct positive and negative examples to guide the training process. Moreover, DSSLP proposes a model-split strategy to accelerate the speed of inference process of the link prediction task. Experimental results demonstrate that the effectiveness and efficiency of DSSLP in serval public datasets as well as real-world datasets of industrial-scale graphs.

Internet of Vehicles (IoV) is the basis of future intelligent transportation systems. Both the control signaling and data dissemination services in IoV must be transmitted with high reliability and low latency so that safety can be guaranteed. Based on the discussion and analyses of issues in achieving high reliability low latency transmissions, we present in this paper a Polar code-based overlapping transmission scheme in which simultaneous transmissions from different nodes to the same receiving node are allowed to use the same time-frequency resource block. To effectively eliminate multiple access interference introduced by the overlapped nonorthogonal transmissions, a successive cancellation list-based improved interference elimination decoding algorithm (SCL-based IIEDA) is proposed to retrieve the Polar coded information. Numerical results show that the SCL-based IIEDA performs well on information recovery in the presented transmission scheme. In addition, it is shown that the proposed scheme not only significantly reduces the acknowledgment overhead but also greatly improves the spectral efficiency.

In wireless sensor networks, the most significant traffic pattern is data gathering from sensor nodes to sink through multihop paths forming a tree structure. This paper presents an energy-efficient MAC protocol based on routing information, EBRI-MAC, designed for such traffic pattern. In EBRI-MAC, an idea of clustering based on data deliveries on the tree is proposed, in which a parent node and its child nodes form a virtual cluster. Based on this structure, a medium access scheme of chain-invitation combined CSMA/CA and a scheme of active/sleep schedule synchronization within each virtual cluster are designed in order to save energy. Considering the major sources of energy waste, the parameter nEis defined to evaluate the energy efficiency, and the theoretical analysis has been performed. By comparison results through ns-2 simulation, it has been verified that EBRI-MAC can win better energy-efficiency than IEEE802.11 DCF and S-MAC under the traffic pattern of each node being time-correlated.